Questions about the knn_graph cosine distance calculation results

[korcys]

Hi, I got some trouble while using torch_clsuter.knn_graph.
When I run this code, it gives the following results.

import torch
from torch_cluster import knn_graph

x = torch.Tensor([[-1, -1], [-1, 1], [1, -1], [1, 1]]).cuda()
batch = torch.tensor([0, 0, 0, 0]).cuda()
edge_index = knn_graph(x, k=2, batch=batch, cosine=True)
print(edge_index)

>>>tensor([[1, 2, 0, 2, 0, 1, 0, 1, 2],
          [0, 0, 1, 1, 2, 2, 3, 3, 3]], device='cuda:0')

Shouldn't cosine <[-1, 1], [1, -1]> equal -1, and cosine <[-1, 1], [1, 1]> equal 0? Why the two nearset points to point 1 are 0, 2 instead of 0, 3.
Besides, why does point 3 have three nearest neighbors instead of two?

Versions:
python 3.7.4
pytorch 1.2.0
torch-cluster 1.4.5

[github-actions]

This issue had no activity for 6 months. It will be closed in 2 weeks unless there is some new activity. Is this issue already resolved?